A deep-learning algorithm could detect earthquakes by filtering out city noise

A deep-learning algorithm could detect earthquakes by filtering out city noise
Cities are loud places. Noise is generated by traffic, trains, and machinery. It can be a nuisance at times, but it can be a problem when it comes time to detect earthquakes. It’s hard to detect the signal of an approaching earthquake in seismic data, especially when there are general vibrations that are typical of cities like New York. This is known as urban seismic noise.

Stanford researchers have discovered a way to get a better signal. The algorithm was described in a paper published in Science Advances today . It is claimed to improve the detection of earthquake monitoring networks in urban areas and other built-up areas. It can improve signal quality and recover signals that were previously too weak to register by filtering out urban seismic sound.

Algorithms trained to sift out urban seismic noise could be of particular use to monitoring stations in and around bustling earthquake-prone cities in South America, Mexico, the Mediterranean, Indonesia, and Japan.

Earthquakes can be monitored by seismic sensors (also known as seismometers), which continuously measure seismic waves in the ground. The Stanford team’s deep-learning algorithm, called UrbanDenoiser, has been trained on data sets of 80,000 samples of urban seismic noise and 33,751 samples that indicate earthquake activity. They were taken in California from both rural San Jacinto and Long Beach.

When applied to the data sets taken from the Long Beach area, the algorithms detected substantially more earthquakes and made it easier to work out how and where they started. And when applied to data from a 2014 earthquake in La Habra, also in California, the team observed four times more seismic detections in the “denoised” data compared with the officially recorded number.

This is not the only AI-based work being done to find earthquakes. Researchers from Penn State have been training deep-learning algorithms to accurately predict how changes in measurements could indicate forthcoming earthquakes–a task that has confounded experts for centuries. Stanford researchers have also trained models for phase picking, which is measuring the arrival times and durations of seismic waves within an earthquake signal. This can be used to determine the location of the quake.

Deep learning algorithms are especially useful for earthquake monitoring as they can take the burden off of human seismologists. Paula Koelemeijer, a Royal Holloway University of London seismologist, was not involved in this research.

In the past, seismologists looked at graphs that were generated by sensors that recorded the motion of ground during earthquakes and would identify patterns by looking at them. Koelemeijer believes deep learning could speed up the process and make it more accurate by allowing for the reduction of large amounts of data.
“Showing that [the algorithm] works in a noisy urban environment is very useful, because noise in urban environments can be a nightmare to deal with, and very challenging,” she says.

Read More